Quote:
Originally Posted by DUR0N
I'm not sure if the Intel NIC's software supports HSRP. Usually they do active load balancing of some kind, or regular failover.
|
HSRP is run on the server providers routers, not the OPs server.
OneLittleBird, what you are looking for is NIC bonding.
- Instructions for CentOS / RedHat:
http://www.cyberciti.biz/tips/linux-...interface.html and some more details
http://www.howtoforge.com/network_card_bonding_centos
-
http://www.cyberciti.biz/howto/quest...iver-howto.php - section 12.2 is basically what you'll be doing - it sounds like you'll be connected to two of their routers - to avoid any potentially confusing situations you may want to run in active-backup mode where only 1 NIC/mac address is active at a time.
If you're running windows, it might get a bit more complicated and specific to the OS and NIC drivers. You mention Intel teaming and Switch Fault Tolerance (SFT) - according to the Intel NIC teaming White Paper, SFT requires STP on your uplink - but you have no control over that. I think what you want is Adapter Fault Dolerance (AFT) which chooses a primary Active NIC and switches to a Standby NIC if the primary one fails.
I recommend that, when using bonding, you should be sure to do ARP/traffic monitoring to detect failures and not just link down monitoring. The Intel whitepaper says 'AFT uses four indicators to detect if a failover is needed: the primary’s link status, the primary’s hardware status, a probe mechanism between the members of the team, and the
primary port’s packet receive counters.' so you should be set in that regard
Best of luck!